Variants of Transitive Inference Reveal the Role of Awareness in Relational Learning

Researchers have long asserted that the flexibility, sensitivity to context, and inference in memory require deliberative processes (e.g. Cohen, Poldrack & Eichenbaum, 2010). Successful performance on the transitive inference task (TI), requires these three attributes, flexibility, context, and inference, for the formation relational association and memories. Numerous studies are in disagreement about the necessity of deliberative processes in relational memory (e.g. Greene, Spellman, Dusek, Eichenbaum, & Levy, 2001). In a typical TI task, participants learn relations among pairs of items, A\u3eB, B\u3eC, C\u3eD, D\u3eE, which if integrated into a hierarchy, A\u3eB\u3eC\u3eD\u3eE, allows correct inferences on pairs that have never been trained such as B\u3eD. The TI task is of particular interest because the manner in which we gather and integrate general world knowledge to make predictions and inferences is modeled by the TI task. It is possible that the subtle differences in design elicit distinct task demands, which result in variable outcomes. This is of theoretical importance because the differences in outcomes arising from design variations have obscured attempts to ascertain whether awareness is necessary for solving the TI task. To test this we conducted three versions of the TI task following methods known to yield distinctly dissimilar outcomes while using a single set of stimuli, with a single population, in the same laboratory. Results from this study suggest that different methodologies, previously assumed to be unimportant, create distinct task demands which yield distinct outcomes. Awareness in these studies appears to be related to the number of stimulus presentations rather than being related to the capacity for successful inference